[sane-devel] brightness/contrast/gamma LUT ideas?

Jim McQuillan jam at McQuil.com
Mon Jul 10 23:55:23 UTC 2006


Allan,

I don't have an answer, but I'd like to comment that I sure could use 
such a feature.

I'm using the Fujitsu fi-5120C scanner, as you know, because you helped 
me get it working.

And so far, it's working really well.  I did find one issue though.  I'm 
scanning both sides of the page at the same time, and i've found that 
documents printed on 20lb paper are getting some "bleed through". That 
is, the scan of the front is actually showing some of the stuff that is 
on the back of the page.

Also, the scans come out looking somewhat "dirty".   Again, because I 
think light from the other side is bleeding through.

I tried printing the same documents on 28lb paper, and the scans look 
beautiful, no bleeding is seen.

This turns out to be a big issue, because when I scan the document 
printed on 20lb paper, the two images tarred together and compressed 
with bzip are taking about 2.7mb.  While the same document printed on 
28lb paper compresses down to about 900kb.  HUGE difference in 
compressability.

I'm thinking that if I could reduce the brightness of the light, it may 
not bleed through so badly.

Thanks,
Jim McQuillan
jam at Ltsp.org




m. allan noah wrote:
> recent model fujitsu scanners dont have native 
> brightness/contrast/gamma support, instead they use an 256x256 or 
> 1024x256 bit look up table to convert the raw scan data before 8 bit 
> output.
>
> while it is true that the 8bit square LUT could be done after scanning 
> with no data loss, most command line front-ends dont do this, and the 
> 10 bit lut has 'access' to more data that never gets out of the 
> scanner, so i would like to extend the backend to provide at least 
> brightness/contrast for these scanners.
>
> i need suggestions or pointers to code that i could use. what things i 
> can find are far over my head, but i have a simple brightness 
> adjustment that shifts the linear slope of the in-out function up or 
> down, and a crude contrast setting that changes the slope of the 
> function around the center of the table. i think both of these methods 
> are likely too simplistic.
>
> anyone?
>
> allan
>



More information about the sane-devel mailing list